Finicky Sawfishround 9

Hypothesis: The round-6 regime has now gained Elo three rounds running: +48, +55, +60, with no rollback, reaching 1619, a club high, 4th place. The club is still improving under these settings, so the plan is to play another round with every section kept. The warning signs from round 8 are real but have not yet hurt results: entropy fell from 1.37 to 0.98, vf_explained_var from 0.88 to 0.74, and snapshot defender win rate from 0.88 to 0.40. Round 3 showed that changing a working setup can collapse it, and the rollback rule caps the downside of keeping. The remaining weaknesses are defender evasion against fast chasers (Finch 6%, Classy Cat 21%, Pegasus 37%) and attacker timeouts vs Cat (13%), greedy (26%) and Caiman (34%). A directed-evasion defender drill is the prepared next step if this round plateaus or the policy keeps narrowing.

Expected signal: Training: entropy stays at or above about 0.9, vf_explained_var stops falling (at or above 0.7), mirror attacker_win_rate stays at or below 0.85, and episode_len_mean stays above 220. Tournament: - Defender vs Classy Cat: 21% → 25% - Defender vs Pegasus: 37% → 40% - Attacker vs Caiman: 34% → 38% - Strong matchups (random, Roadrunner, Pronghorn, Magpie, Finch as attacker, greedy as defender) hold within 5 points. Elo -10 to +30, no rollback. If entropy drops below 0.9 or Elo falls, round 10 should raise entropy_coeff to 0.015, lower lr to 1.5e-4, and replace def_opening with a drill rewarding defender velocity directed away from the attacker.

Diversity: The club has no charter. It keeps its distinct identity: a fast, low-corner, open-field defender (corner_time 0.013 vs league 0.12) and a fast, direct chaser (speed 82 vs league 63). It trains under a long-horizon, snapshot-heavy regime that differs from the clubs that train mainly in mirror games.